Query Best Practices for the Reporting Agent
As with any AI system, the quality of the Reporting Agent's results depends on the clarity and structure of the instructions it receives.
Well written prompts reduce ambiguity, help the Agent understand the user's intent, and produce more accurate, relevant, and consistent responses. Clear prompts also save time by minimizing the need for repeated corrections or follow-up questions.
Therefore, we provide in this article some basic principles to support you in engaging successfully with JedoxAI and unlocking the full potential of the Reporting Agent.
Tips and examples for better prompts
1. Be specific
Vague queries produce vague answers. Specify the exact metric, dimension, and time period.
| Example 1 | Metric clarity |
Less effective  |
Show me sales data |
Good!  |
Show me total revenue by product line for Q1 2024 |
| Example 2 | Time clarity |
| Less effective |
Show me strategic |
| Good! |
Show me revenue for Strategic version by region in Q1 2024 |
2. Break down complexity
One focused question works better than asking three things at once. The Agent can then focus properly on one topic, then build on top of it.
Iterative queries lead to better results.
| Example | |
| Less effective |
Show me revenue trends, top customers, and forecast accuracy by region for the last 6 months |
| Good! |
Step 1: Show me revenue trends by region for the last 6 months Step 2: Now show me the top 5 customers in the region with highest growth |
3. Correct and refine
When results aren’t quite right, use follow-up questions to guide the Agent.
The Reporting Agent is conversational by design - treat it like a colleague!
Example:
-
Let's say you get an incorrect result.
-
Clarify: "Actually, show me the actual not budget".
-
The Agent corrects and responds accurately.
-
You refine it: "Now break this down by product".
Updated May 12, 2026